Gordon Alexander, Visiting Professor of Finance
CAMBRIDGE, Mass., February 26, 2010 — The current risk management system for international banks is insufficient to prevent large losses of capital, according to MIT Sloan School of Management Visiting Professor Gordon Alexander. In his research on the current regulatory framework for trading books – which utilizes Value-at-Risk and Stress Testing to assess downside risk and to determine capital adequacy – he found those measures to be inadequate to prevent future economic turmoil.
“Regulators are worried about systemic risk,” said Alexander. “The concern is that if one of the big banks goes bankrupt, it will trigger a massive wave of defaults and bankruptcies that will disrupt the economy on a large scale and cause a global recession that could dwarf what we have now. We want to minimize the chance of future global crises, but the current risk measures required by the Basle Committee on Banking Supervision are not enough.”
In his paper, “Bank Regulation, Risk Management, and Financial Stability,” Alexander studies the effectiveness of Value-at-Risk and Stress Testing in controlling downside risk for banks’ trading books, which involve interest rate related instruments and equities such as government securities, corporate bonds, and common stocks. The current regulatory framework, he said, requires international banks to use Stress Testing to assess how their portfolios would be affected if events such as 9/11 happened again. Additionally, banks are required to maintain a certain amount of equity within the organization at all times. To determine that amount, banks use the other required measure, Value-at-Risk (VaR). That measure assesses the size of losses over the next 10 days under “normal conditions,” which are defined as conditions that occur 99% of the time.
“For example, a bank might state that it is 99% sure that it will not lose more than $1 billion dollars over the next 10 days,” explained Alexander. “If so, the bank would be required to hold between $3 and $4 billion in equity capital in support of the trading portfolio, with the exact amount depending on other factors.”
The problem, he noted, is that VaR only measures the “lower bound” of possible losses whereas bank losses have historically been surprisingly larger than this bound. Continuing with his example, “a $1 billion VaR is equivalent to saying that there is a 1% chance a bank can suffer a loss of at least $1 billion over the next 10 days, but it would be also useful to know that the expected loss for the bank, should one occur, is $5 billion – the Conditional Value-at-Risk (CVaR),” he said.
“CVaR is more difficult to measure with a high degree of precision, but just using VaR won’t prevent the next great recession. We’re better off thinking about the kinds of downside risk measures like CVaR than ignoring them completely because we want banks prepared to weather storms,” said Alexander.
He further found that whether or not a bank allows short selling is a significant factor in the effectiveness of using VaR and Stress Testing to control downside risk. “If banks are not involved in short selling then these measures work reasonably well, but when banks are involved in short selling – and almost all of the big ones are -- then these measures are not that helpful in controlling downside risk for trading books,” he explained.
“Our results support the view that risk management systems based on the use of historical simulation to measure VaR and conduct Stress Tests are of dubious effectiveness in controlling downside risk. Hence, the Basle framework is of questionable adequacy in promoting financial stability,” wrote Alexander. The paper was coauthored by Alexandre M. Baptista of The George Washington University, and Shu Yan of the University of South Carolina.